Efficiency multi-level distribution method for virtual power plant participating in multi-element coupling market

Abstract

An efficiency multi-level distribution method for a virtual power plant participating in a multi-element coupling market comprises the following steps: firstly, establishing an income and risk quantification model of the multi-element market to obtain income and risk under a multi-element market coupling condition; then, solving an optimal investment portfolio which considers income and risk at the same time by adopting an investment portfolio method, and formulating an optimal bidding strategy which participates in the multi-element coupling market for different virtual power plants; and finally, establishing an income distribution model considering output and risk, and performing efficiency distribution on distributed resources in the virtual power plant in combination with an improved Shapley value method. By means of the method, reasonable income distribution of the distributed resources and the supply capacity of the power system can be guaranteed, and the distributed resources can be promoted to improve the accuracy of predicting self output and the enthusiasm of participating in the market.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy power supply, and specifically relates to a multi-level allocation method for the efficiency of a virtual power plant participating in a multi-coupling market. Background Art

[0002] The introduction of the concept of virtual power plant (VPP) and its gradual application have enabled a large number of distributed energy resources (DERs) to participate more widely and deeply in the power supply market. This process aims to give full play to the potential flexibility of DERs and provide a variety of services for the power grid, covering energy supply, frequency regulation, reactive power compensation and many other aspects.

[0003] However, DERs show significant uncertainty in output, their demand is also unstable, and market prices fluctuate frequently and violently. This series of factors makes it difficult for VPPs to accurately and reasonably plan their output in multiple markets within the existing market mechanism framework. This situation also poses a serious constraint on the economy and flexibility of VPP regulation, becoming a key bottleneck hindering its development and optimization.

[0004] In the current technical system, calculating the optimal unit combination is considered one of the main ways to solve the above problems. This model regards VPP as a passive recipient of market prices. Although it has achieved an increase in the VPP's profit level to a certain extent, it has ignored the detailed trade-off between expected returns and potential risks to a considerable extent. In view of this, it is particularly important to comprehensively and in-depth consider the close coupling between different markets and the inherent logical correlation of returns. Only according to the specific risk situation faced by different markets can the most appropriate virtual power plant output allocation mode be accurately selected, thereby achieving optimal allocation of resources and effective risk control.

[0005] In addition, since the DERs (single distributed resources) aggregated by VPPs usually belong to completely different interest groups, the reasonable distribution of benefits has become a key element to achieve a win-win situation for all parties and enhance the VPP's ability to supply the power system. In current practical operations, the commonly used benefit distribution methods mainly include the kernel method and the Shapley value method. However, such methods have exposed significant defects in actual application. They fail to fully take into account the ability to ensure the stable supply of the power system while increasing the benefits of DERs. Once the actual output of DERs deviates from the preset control signal and excessively pursues the maximization of their own interests, it is very likely to induce a series of serious problems, such as a serious imbalance in power supply and demand and a sharp decline in power grid stability.

[0006] It can be seen that in order to effectively deal with market risks and successfully avoid the above-mentioned difficult problems, existing research has actively introduced the key concept of conditional risk value. With the pursuit of maximum benefits as a clear goal, a risk-averse scheduling optimization model for VPP is constructed. Although such research provides a corresponding theoretical basis for the operation strategy of VPP, it can effectively avoid risks and significantly improve the level of benefits to a certain extent, but it has failed to successfully build a scientific, reasonable and complete benefit distribution mechanism. This makes it impossible to fully ensure the fairness and rationality of the benefit distribution of various subjects, which has an extremely adverse impact on the establishment of a long-term, stable, mutually beneficial and win-win and solid and lasting cooperative relationship between DERs.

[0007] In summary, the existing efficiency (including output and revenue) distribution needs to be improved to provide a DERs revenue distribution mechanism that takes output deviation and risk control into account. Summary of the invention

[0008] The present invention provides a multi-level allocation method for the efficiency of a virtual power plant participating in a multi-coupling market to solve the above-mentioned problems.

[0009] The present invention adopts the following technical solutions:

[0010] A multi-level allocation method for the efficiency of a virtual power plant participating in a multi-coupling market, the main steps of which include:

[0011] Establish a multi-market model based on the power supply network with participation of virtual power plants;

[0012] The multi-market model includes at least an energy arbitrage market model and a distributed network operator market model, and calculates the expected return and risk prediction value of a single participant in the multi-market model;

[0013] According to the portfolio method, the combined expected return of the multi-market investment portfolio is calculated by the expected return, and the combined risk value of the multi-market investment portfolio is calculated by the risk prediction value;

[0014] Based on the risk and return optimization model, the expected return of the portfolio and the risk value of the portfolio are brought in to obtain the optimal investment portfolio; the optimal bidding strategy is obtained through the optimal investment portfolio;

[0015] Obtaining a clearing result of the virtual power plant based on the optimal bidding strategy;

[0016] Establish a multi-level efficiency distribution model, which at least includes an output distribution model and a benefit distribution model;

[0017] The output allocation model allocates the clearing results of the virtual power plant to the distributed energy cluster, and calculates the optimal output allocation strategy of the virtual power plant to the distributed energy cluster through a dynamic stochastic control model and a mean field control framework;

[0018] The profit distribution model distributes the profit of the virtual power plant based on the Shapley value method, the risk value method and the optimal output allocation strategy.

[0019] Optionally, the expected income of a single participant in the energy arbitrage market model is the sum of the expected unit income multiplied by the number of arbitrage units in the settlement period, and the formula is:

[0020]

[0021] In the formula, R E is the expected return of a single participant, EAM is the number of arbitrage units, r EAM,t is the expected unit revenue; n is the number of settlement cycles, and t is the settlement cycle;

[0022] The formula for the expected unit revenue is:

[0023]

[0024] In the formula, is the price of electricity within the settlement period t, is the purchase price of electric power in the settlement period t;

[0025] The risk prediction value of a single participant in the energy arbitrage market model is the standard deviation of the predicted value of each energy sale price, and the formula is:

[0026]

[0027] In the formula, σ EAM,t is the risk prediction value of a single participant in the energy arbitrage market model, is the xth of the n predicted values ​​of the selling price in settlement period t.

[0028] Optionally, the distributed network operator market model establishes branches and nodes based on congestion management technology;

[0029] The expected profit of a single participant in the distributed network operator market model is the sum of the expected unit price difference multiplied by the number of transaction units during the settlement period, and the formula is:

[0030]

[0031] In the formula, R D is the expected return of a single participant, DNO is the number of trading units, rDNO,t is the expected unit price difference; n is the number of settlement cycles, and t is the settlement cycle;

[0032] The formula for the expected unit price difference is:

[0033]

[0034] In the formula, CC l,t is the congestion cost of branch l in settlement period t, the cost of purchasing energy for the distributed network operator market;

[0035] The risk prediction value of a single participant in the distributed network operator market model is the standard deviation of the congestion volume in each time period, and the formula is:

[0036]

[0037] In the formula, σ DNO,t is the risk prediction value of a single participant in the distributed network operator market model, CC i,t,z is the zth of the h predicted values ​​of the congestion cost due to load changes in settlement period t.

[0038] Optionally, the multi-market model further includes a frequency modulation market model;

[0039] The expected income of a single participant in the frequency modulation market model is the sum of the expected income of the frequency modulation service unit in the settlement period multiplied by the number of frequency modulation service units, and the formula is:

[0040]

[0041] In the formula, R F is the expected revenue of a single participant in the frequency modulation market model, FRM is the number of frequency modulation service units, r FRM,t is the expected revenue of the frequency modulation service unit; n is the number of settlement cycles, and t is the settlement cycle;

[0042] The expected revenue per frequency regulation service unit is the unit price of the frequency regulation service and the unit energy cost of the frequency regulation service, and the formula is:

[0043]

[0044] Where, REF t is the unit price of frequency modulation service, Unit energy cost for frequency regulation services.

[0045] Optionally, the process of calculating the unit price of frequency modulation service of a single participant in the frequency modulation market model includes:

[0046] Calculate capacity benefits and capacity response benefits;

[0047] The capacity benefit is calculated as:

[0048] AF=LF+HF=LFR×LA+HFR×HA;

[0049] Wherein, AF is the capacity benefit of a single participant, LF is the adjustable capacity fee, HF is the adjustable capacity fee, LFR is the adjustable capacity unit fee, HFR is the adjustable capacity unit fee, LA is the adjustable capacity of the virtual power plant within the time period, and HA is the adjustable capacity of the virtual power plant within the time period;

[0050] The formula for the capacity response benefit is:

[0051] RE t =CAP t ×RP t ;

[0052] In the formula, RE t is the capacity response revenue in settlement period t, CAP t is the response capacity of settlement period t, RP t Pay a price for the unit response;

[0053] The formula for the unit response payment price is:

[0054]

[0055] Where, E is the number of capacity frequency modulation, MIP e,t MIV is the market price index for the e-th capacity modulation in settlement period t. e,t is the market volume index for the e-th capacity adjustment in settlement period t; FX t To pay the price floor;

[0056] The formula for the unit price of the frequency modulation service is:

[0057] REF t =AF+RE t =AF+CAP t ×RP t ;

[0058] Where, REF t The unit price for frequency modulation service.

[0059] Optionally, the risk prediction value of a single participant in the frequency modulation market model is the standard deviation of the energy price when the virtual power plant purchases energy from the system, and the formula is:

[0060]

[0061] In the formula, σ FRM,t is the risk prediction value of a single participant in the FM market model, is the yth of m predicted values ​​of the energy price for frequency modulation in settlement period t.

[0062] Optionally, the expected return of the portfolio is the sum of the weighted profitability of each market share, calculated as follows:

[0063]

[0064] In the formula, R mp,t is the expected return of the portfolio, ω i is the weight of the virtual power plant output in each market in the portfolio, r i,t is the unit revenue of the virtual power plant in the settlement period t of the i-th market, P VPP,t is the total output of the virtual power plant in settlement period t, R i,t is the revenue of the virtual power plant when the output is 100% in the settlement period t of the i-th market; M is the market set;

[0065] The portfolio risk value is calculated by the sum of the risk prediction values ​​of the single participants of each market share in the portfolio and the correlation between any two markets, and the formula is:

[0066]

[0067] In the formula, σ mp,t is the portfolio risk value in settlement period t, ω i is the weight of the virtual power plant in the portfolio in the i-th market, ω j is the weight of the virtual power plant in the portfolio in the jth market; σ i,t is the risk prediction value of the ith market in settlement period t, σ j,t is the risk prediction value of the jth market in settlement period t; ρ ij is the correlation coefficient between the ith market and the jth market.

[0068] Optionally, the calculation process of the risk and return optimization model includes:

[0069] Establish risk minimization models and profit maximization models;

[0070] Establishing a utility function based on the results of the risk minimization model and the benefit maximization model;

[0071] The utility function is solved by Lagrangian function, and the optimal investment portfolio is obtained.

[0072] Optionally, the output distribution model calculation process includes:

[0073] Establishing a set of participants;

[0074] Construct decision vector space;

[0075] Establishing a utility function based on the set of participants and the decision vector space, wherein the utility function includes a preference degree parameter in the interaction of the participants;

[0076] Establishing the dynamic random control model through the utility function and incentive strategy;

[0077] The mean field control framework is established based on the dynamic random control model, and the optimal output allocation strategy is solved.

[0078] Optionally, the profit distribution model calculation process includes:

[0079] Establish the profit sharing function of the participants based on the Shapley value method;

[0080] Based on the risk utility of distributed resources, the output deviation of distributed power sources and the revenue contribution of operators, the revenue sharing function is optimized to obtain an improved revenue sharing function;

[0081] The profit distribution strategy of the virtual power plant is obtained by solving the improved profit sharing function.

[0082] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0083] The present invention establishes a quantitative model of benefits and risks in multiple markets, considers the optimal investment portfolio of benefits and risks, and formulates the optimal bidding strategy; combines the benefit distribution model considering output and risk, distributes benefits to DERs in the VPP, and can ensure the reasonable distribution of benefits of DERs and the supply capacity of the power system, and can promote DERs to improve the accuracy of predicting their own output and the enthusiasm of participating in the market. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0085] Figure 1 is a flow chart of the method of the present invention;

[0086] Figure 2 is a graph of the benefits and risks of the virtual power plant participating in the optimal investment portfolio of two markets in Example 1 of the present invention;

[0087] Figure 3 is a graph of the benefits and risks of the virtual power plant participating in the optimal investment portfolio of three markets in Example 1 of the present invention;

[0088] Figure 4 is the output curve of different distributed resources in Example 1 of the present invention, Figure 4 Figure (a) is the photovoltaic output curve. Figure 4 Figure (b) is the wind output curve. Figure 4 Figure (c) shows the building cooling and heating load curve under temperature control conditions. Figure 4 Figure (d) is the distributed energy storage output curve. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0091] It should also be noted that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products unless otherwise specified.

[0092] The present invention provides a multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market, such as Figure 1 As shown, the main steps include:

[0093] 1. Establish a multi-market model based on the power supply network in which virtual power plants (hereinafter referred to as VPPs) participate, and calculate the expected returns and risk prediction values ​​of single participants in the multi-market model.

[0094] In this embodiment, the multi-market model includes an energy arbitrage market model, a frequency modulation market model and a distributed network operator market model.

[0095] 1.1 Energy Arbitrage Market (EAM) Model;

[0096] Alternatively, price arbitrage in the energy arbitrage market model means that VPP acts as virtual energy storage to charge during low electricity price periods and discharge during high price periods, but this requires a larger price difference to ensure that the initial investment can be paid back. However, the uncertainty of energy arbitrage comes from price changes, which cannot be accurately predicted.

[0097] Therefore, the expected profit of a single participant in the virtual power plant market is the sum of the expected unit profit multiplied by the number of arbitrage units in the settlement period, and the formula is:

[0098]

[0099] In the formula, R E is the expected return of a single participant, EAM is the number of arbitrage units, r EAM,t is the expected unit revenue; n is the number of settlement cycles, and t is the settlement cycle.

[0100] Furthermore, the formula for expected unit revenue is:

[0101]

[0102] In the formula, is the electricity price during the settlement period, The purchase price of electric power during the settlement period;

[0103] The risk prediction value of a single participant in the energy arbitrage market model is the standard deviation of the predicted value of each energy sale price, and the formula is:

[0104]

[0105] In the formula, σ EAM,t is the risk prediction value of a single participant in the energy arbitrage market model, is the xth of the n predicted values ​​of the selling price in settlement period t.

[0106] 1.2 Frequency Regulation Market (FRM) Model;

[0107] The expected revenue of a single participant in the FRM model is the sum of the expected revenue of the frequency regulation service unit in the settlement period multiplied by the number of frequency regulation service units. The formula is:

[0108]

[0109] In the formula, R F is the expected revenue of a single participant in the frequency modulation market model, FRM is the number of frequency modulation service units, r FRM,t is the expected revenue of the frequency modulation service unit; n is the number of settlement cycles, and t is the settlement cycle.

[0110] Among them, the expected revenue of frequency regulation service unit is the unit price of frequency regulation service and the unit energy cost of frequency regulation service, and the formula is:

[0111]

[0112] Where, REF t is the unit price of frequency modulation service, Unit energy cost for frequency regulation services.

[0113] The calculation process of the unit price of frequency regulation service of a single participant in the frequency regulation market model includes: FRM is open to all third-party independent entities in the power system with an installed capacity of more than 5MW through a competitive bidding process. Distributed resources participate in FRM through VPP aggregation, and the frequency regulation benefits obtained usually include two types of fees: capacity benefits and capacity response benefits. These two benefits are mainly issued in the form of separate payments, which can be specifically capacity compensation fees and capacity response fees.

[0114] From this, the capacity benefit and capacity response benefit are calculated;

[0115] When the frequency response VPP is available at any time, the market provides capacity compensation fees. Regardless of whether a response is required within a period of time, the market will pay this fee, so the formula for capacity benefit AF is:

[0116] AF=LF+HF=LFR×LA+HFR×HA(6)

[0117] In the formula, AF is the capacity benefit of a single participant, LF is the adjustable capacity fee, HF is the adjustable capacity fee, LFR is the adjustable capacity unit fee, HFR is the adjustable capacity unit fee, LA is the adjustable capacity of VPP in the time period, and HA is the adjustable capacity of VPP in the time period;

[0118] When VPP responds to frequency regulation instructions, the market will pay capacity response fees. VPP can provide two responses: charging when the frequency needs to be lowered and discharging when the frequency needs to be raised. The specific charging and discharging actions depend on the state of charge. The response payment price is designed as follows: increasing output pays a higher fee, and reducing output pays a lower fee. Therefore, the formula for capacity response benefits is:

[0119] RE t =CAP t ×RP t (7)

[0120] In the formula, RE t is the capacity response revenue in settlement period t, CAP t is the response capacity of settlement period t, RP tPrice paid for unit response.

[0121] The formula for the unit response payment price is:

[0122]

[0123] Where, E is the number of capacity frequency modulation, MIP e,t is the market price index of the eth time in settlement period t, MIV e,t is the market volume index of the eth time in settlement period t; FX t The payment price floor is a floating design. When the frequency needs to be increased, the FX t for FX when the frequency needs to be lowered t It is 0.75￡ / MWh.

[0124] Combining equations (7) and (8), the formula for the unit price of frequency modulation service is:

[0125] REF t =AF+RE t =AF+CAP t ×RP t (9)

[0126] Where, REF t The unit price for frequency modulation service.

[0127] Furthermore, the unit expected benefit of frequency regulation comes from the difference between the unit price of providing frequency regulation services and the unit energy cost of frequency regulation; since the unit price of frequency regulation is constant, the risk comes from the fluctuation of energy prices when VPP purchases energy from the system, that is, the risk prediction value of a single participant in the frequency regulation market model is the standard deviation of the energy price when VPP purchases energy from the system, and the formula is:

[0128]

[0129] In the formula, σ FRM,t is the risk prediction value of a single participant in the FM market model, is the yth of m predicted values ​​of the energy price for frequency modulation in settlement period t.

[0130] 1.3 Distributed Network Operator's Market (DNO's market, hereinafter referred to as: DNO) model;

[0131] The introduction of VPP can reduce the peak load of the power grid. By providing appropriate congestion management services to distributed network operators, VPP can help save network investment, operation and maintenance costs. In this market, the risk comes from unexpected network overload due to uncertain demand and power generation. Therefore, VPP is needed to establish branches and nodes based on congestion management technology in the distributed network operator market model.

[0132] Among them, taking branch l as an example, its congestion cost CC in settlement cycle t is l,t for:

[0133]

[0134] In the formula, ΔP l,t is the power flow of branch l due to constraint changes, P wl,t is the unconstrained branch l power flow, P nl,t is the branch power flow with constraints, TCC t The total cost of the system changes.

[0135] Furthermore, we solve the congestion cost. Assuming that the number of branches associated with node o in the system is L, the congestion cost change of node o in settlement period t is TC o,t It is expressed as follows,

[0136]

[0137] The impact of node demand / generation changes on branch flows can be quantified by the power transfer factor PTDF. o1o2,l It represents the proportion of energy transferred from one node to another through branch l. It is the sensitivity matrix of the change of active power flow of the line caused by the change of node power. Ignoring the line loss, based on the reactance of the transmission line, the DC power flow model is used to determine the proportion of power flow through the line (o1o2, l).

[0138]

[0139] Where, X o1o1 and X o2o2 They refer to the self-impedance of node o1 and node o2 respectively; X o1o2 Refers to the mutual impedance between node o1 and node o2.

[0140] Therefore, the demand / generation change ΔP at node o1 o1,t Changes in branch power flow ΔP l,t As follows:

[0141] ΔP l,t =PTDF o1o2,l ×ΔP o1,t (14)

[0142] The expected return of a single participant in the DNO model is the sum of the expected unit price difference multiplied by the number of trading units during the settlement period, and the formula is:

[0143]

[0144] In the formula, R D is the expected return of a single participant, DNO is the number of trading units, r DNO,t is the expected unit price difference; n is the number of settlement cycles, and t is the settlement cycle.

[0145] The expected return of the DNO is determined by the difference between the expected energy sales price and the energy purchase cost in the DNO market. The expected energy sales price is the congestion cost saved due to the operation of the VPP. If there is no congestion in the system, the expected return of this market is zero. Therefore, the formula for the expected unit price difference is:

[0146]

[0147] In the formula, CC l,t is the congestion cost of branch l in settlement period t, Energy purchasing costs for the distributed network operator market.

[0148] The risk of DNO is the standard deviation of congestion in each time period (i.e., the difference between each possible congestion cost and the average congestion cost). Therefore, the risk prediction value of a single participant in the DNO model is the standard deviation of congestion in each time period, and the formula is:

[0149]

[0150] In the formula, σ DNO,t is the risk prediction value of a single participant in the distributed network operator market model, CC i,t,z is the zth of the h predicted values ​​of the congestion cost due to load changes in settlement period t.

[0151] 2. According to the portfolio method, the expected return of the diversified market investment portfolio is calculated by the expected return, and the risk value of the diversified market investment portfolio is calculated by the risk prediction value.

[0152] Among them, the expected return of the portfolio is the sum of the weighted profitability of each market share, and the formula is:

[0153]

[0154] In the formula, R mp,t is the expected return of the portfolio, ω i is the weight of VPP's contribution in each market in the investment portfolio, ri,t is the unit income of VPP in the settlement cycle t of the i-th market, P VPP,t is the total output of VPP in settlement period t, R i,t is the profit of VPP when the output is 100% in the settlement period t of the i-th market; M is the market set.

[0155] The portfolio risk value is calculated by the sum of the risk prediction values ​​of the single participants in each market share of the portfolio and the correlation between any two markets. The formula is:

[0156]

[0157] In the formula, σ mp,t is the portfolio risk value in settlement period t, ω i is the weight of VPP's output in the i-th market in the portfolio, ω j is the weight of VPP's contribution in the jth market in the portfolio; i,t is the risk prediction value of the ith market in settlement period t, σ j,t is the risk prediction value of the jth market in settlement period t; ρ ij is the correlation coefficient between the ith market and the jth market.

[0158] 3. Based on the risk and return optimization model, the expected return and risk value of the portfolio are input to obtain the optimal investment portfolio; the optimal bidding strategy is obtained through the optimal investment portfolio.

[0159] The calculation process of the risk and return optimization model includes:

[0160] 3.1 Establish risk minimization model and profit maximization model;

[0161] The optimization model for risk minimization is:

[0162]

[0163] Where T is the total number of time intervals, R e,t is the expected return of the investment portfolio in settlement period t.

[0164] The portfolio point with the lowest risk among all portfolios is called the global minimum variance portfolio point. It can be determined by the partial derivative of each market share weight, as follows:

[0165]

[0166] In order to jointly consider risks and benefits, this embodiment does not establish a profit maximization model alone, but achieves the goal of pursuing profit maximization by optimizing and solving the utility function. The specific operations are as follows.

[0167] 3.2 Based on the results of the risk minimization model and the benefit maximization model, establish the utility function U t ;

[0168] In order to simultaneously determine the optimal investment portfolio with the lowest risk and highest return of VPP, the utility function of expected return and return variance is established based on formula (20), which is:

[0169]

[0170] Where A is the investor's risk aversion index.

[0171] 3.3 Solve the utility function through the Lagrangian function and obtain the optimal investment portfolio;

[0172] The optimal investment portfolio of the target in formula (22) under the constraints of formula (20) can be determined by the Lagrangian function, which is:

[0173]

[0174] Where Z is the solution of the utility function. Substituting equation (18) and equation (19) into equation (23), we can get the following equation:

[0175]

[0176] By partial differentiation of each variable, the optimal investment portfolio with the highest return can be calculated. Furthermore, the corresponding optimal bidding strategy is obtained based on the optimal investment portfolio, thereby obtaining the clearing result corresponding to the VPP.

[0177] 4. Establish a multi-level efficiency distribution model, which shall at least include an output distribution model and a benefit distribution model.

[0178] 4.1 Output distribution model;

[0179] The output allocation model allocates the clearing results of VPP to distributed energy clusters, and calculates the optimal output allocation strategy of VPP to distributed energy clusters through the dynamic stochastic control model and the mean field control framework. That is, the clearing results of VPP in the multi-market should be further allocated to DER clusters. While pursuing the maximization of their own interests, DER clusters are obliged to undertake the output tasks allocated by VPP. Furthermore, in order to seek the optimal distribution of instructions and economic benefits, interactions are formed between DER clusters, which can be expressed as a dynamic stochastic control model.

[0180] The output distribution model calculation process includes:

[0181] Create a set of participants Represents the number of DER clusters participating in the settlement period t;

[0182] Constructing the decision vector space Represents the joint control strategy of all clusters, where cluster A d The control strategy in settlement period t includes the EAM, FRM and DNO's market output allocated to the DER cluster, i.e.

[0183] Establishing a utility function based on the set of participants and the decision vector space And the utility function includes the preference parameters of the participants in the interaction. d While pursuing the maximization of benefits, it is necessary to respond to the control instructions of VPP. The utility function is defined as the net benefit, and the formula is:

[0184]

[0185] In the formula, s -d It is a DER cluster except A d The joint decision vector of and Complete the unit incentive price for EAM, FRM and DNO's market allocated output on behalf of DER clusters; and Representative A d EAM, FRM and DNO's market allocation output; δ EAM ,δ FRM and δ DNO Deviation from standard output and The penalty coefficient.

[0186] Considering DER cluster A d With similar response characteristics, and The clearing power can be calculated as follows:

[0187]

[0188] In the formula, ω, as mentioned above, specifically represents the output weight of each market, ω EAM ,ω FRM ,ω DNO are the weights of VPP in EAM, FRM and DNO output respectively.

[0189] A dynamic random control model is established through utility function and incentive strategy; according to the regulation capability and frequency regulation requirements of the DER cluster, the dynamic incentive mechanism formula is formulated as follows:

[0190]

[0191] In the formula, and is a piecewise linear function describing the relationship between the adjustment amount and the incentive price; and A d The linear parameter of the adjustment amount and adjustment positivity; EAM , Π FRM and Π DNO is the total amount of upward and downward incentives of the DER cluster as a proportion of the market clearing price; and is the market demand in settlement period t.

[0192] Under the incentive mechanism of formula (27), A d The optimal control strategy to maximize the frequency modulation benefit can be sought based on its own state and frequency modulation incentive price. Therefore, the dynamic stochastic control problem is defined as P1:

[0193]

[0194] In the formula, It is the optimal control strategy at the DSG equilibrium point.

[0195] According to Bellman's optimal value principle, the value function v of problem P1 d is defined as follows, and the HJB equation of P1 is further calculated as:

[0196]

[0197]

[0198] In the formula, It can be obtained by setting the left side of equation (30) to 0.

[0199] Based on the dynamic random control model, a mean field control framework is established to solve the optimal output allocation strategy. To achieve the optimal strategy point of (29), all DER clusters need to interact in pairs. Therefore, as the number of clusters increases, searching for equilibrium points becomes an NP-hard problem. In order to overcome this problem, a mean field control framework is constructed to approximate the solution. In the mean field control framework, DER clusters are no longer interacting in pairs, but rather the interaction between a single participant and the mean field formed by the remaining N-1 participants.

[0200] When large-scale clusters are involved, the impact of a single cluster’s decision on other clusters becomes negligible. However, the impact of the mean field on the decision of each cluster is significant, which can be simulated as a collective effect, namely MFT, which is calculated as:

[0201]

[0202] In the formula, and They refer to the mean field terms corresponding to the cluster output in EAM, FRM and DNO, respectively.

[0203] Combining the incentive mechanism in equation (31) and the corresponding equation (27), the relationship between the incentive price and MFT can be expressed as:

[0204]

[0205] Therefore, J in formula (25) u,d It can be further expressed as:

[0206]

[0207] The MFT in equation (31) can be used to transform the dynamic random control problem into a mean field control problem. The corresponding optimal control strategy is It can be restated through MFT as:

[0208]

[0209] In the mean field control problem, the influence of the mean field on the control strategy of the participants is reflected by the HJB equation. Combining (30) and (31), the HJB equation in mean field control can be obtained as:

[0210]

[0211] Equation (31) describes the mapping from the control strategy of a single homogeneous particle to the average field effect, denoted as the operator Equation (32) describes the mapping from the average field effect to the individual control strategy, denoted as operator The interactive iteration between equations (31) and (35) can eventually reach the mean field control optimal point, thereby obtaining the optimal output allocation strategy of VPP to DER clusters.

[0212] 4.2 Profit distribution model;

[0213] The profit distribution model distributes the profits of VPP based on the Shapley value method (hereinafter referred to as the Shapley value method), the risk value method and the optimal output allocation strategy.

[0214] The calculation process of the income distribution model includes:

[0215] The benefit sharing function of the participants is established based on the Shapley value method. The Shapley value method takes into account the risk utility of DERs, the output deviation of distributed power sources (hereinafter referred to as DG), and the impact of the operator's revenue contribution, and redistributes the benefits of VPP participating in the multi-coupling market among distributed resources. Since DG has output deviation, it is also necessary to consider the economic losses that may be caused by the existing prediction errors; therefore, corrections should be made when designing the internal benefit function of VPP, so the DERs benefit R DERk It includes two parts: the operator’s contract revenue R DERk,s and the risk-return R caused by uncertainty DERk,β , so the revenue function of DERs is:

[0216] R DERk =R DERk,s +R DERk,β (36)

[0217] Based on the risk utility of distributed resources, the output deviation of distributed power sources and the revenue contribution of operators, the revenue sharing function is optimized to obtain an improved revenue sharing function; first, the Shapley value method is to share the revenue of participants according to their marginal contributions, and the formula is:

[0218]

[0219] In the formula, R DERk,s is the apportioned income of entity DERk, P DERk The contract contribution of the subject DERk, R e The combined expected return of a portfolio.

[0220] In addition, due to the different characteristics of DERs, their risk tolerance and contribution levels to VPP are also different. However, the Shapley method assumes that participants have the same risk tolerance, so it also needs to be modified according to the risk utility of DERs.

[0221] The risk-return allocation mechanism is established based on the conditional risk value theory, and the formula is:

[0222] Pr(R DERk,β ≤VaR)=λ(38)

[0223] Wherein, Value at Risk (VaR) represents the value of the risk status in market fluctuations, Pr is the probability that the portfolio value loss is less than the expected maximum possible loss; λ is a given confidence level, which represents the decision orientation of investors.

[0224] According to the conditional risk value theory, the uncertainty loss function can be established The formula is:

[0225]

[0226] Where p1 is the unit penalty factor for output deviation; DERk’s output deviation ΔP DERk The formula is:

[0227]

[0228] In the formula, The actual output of DERk.

[0229] If the output of DERk deviates, the maximum penalty cost is C, then the loss function The probability of not exceeding the threshold C and the risk value function are shown in equations (40) and (41):

[0230]

[0231] In the formula, ψ(P wt ,C) is the loss function with respect to the variable The cumulative distribution function of .

[0232] Therefore, the risk-return R of DERk due to uncertainty DERk,β The formula is:

[0233]

[0234] Therefore, the improved revenue sharing function is:

[0235]

[0236] By improving the revenue sharing function R DERk Solve and obtain the corresponding VPP revenue distribution strategy.

[0237] In Example 1, two cases of VPP participating in two markets and three markets are selected for demonstration.

[0238] When VPP participates in two markets, the optimal investment portfolio of VPPA and VPPB when participating in two markets is determined according to the Lagrangian function of equations (23) and (24): Figure 2 As shown in the figure. For VPP A, its optimal point is that the energy arbitrage market accounts for 70% and the frequency regulation market accounts for 30%. In this period, the expected profit of this VPP is 0.66 yuan. VPP B puts 80% of its capacity into the energy arbitrage market and 20% of its capacity into the frequency regulation market, with an expected profit of 0.91 yuan.

[0239] When VPP participates in three markets, the portfolios of VPPs with different risk aversion levels are as follows: Figure 3As shown. The capacity of VPP B in the energy arbitrage market is 30%, and the capacity in the frequency regulation market is 70%. Since the risk value of VPP A is σA=0.46 at this time, if the expected return is still 0.66 yuan, 45% of the capacity should be invested in the frequency regulation market, 35% of the capacity in the energy arbitrage market, and 25% of the capacity in the DNO market. The best investment portfolio of VPP A is to put 35% of the capacity in the frequency regulation market and 65% in the energy arbitrage market, so that a higher return of 0.74 yuan can be obtained.

[0240] The specific output of heterogeneous DERs is as follows: Figure 4 As shown. Taking the period of high electricity prices as an example, photovoltaic and wind turbines will increase power generation to meet the peak load within the upward power range. During the period of high electricity prices, the temperature control load reduces the cooling capacity by raising the room temperature in advance, and the distributed energy storage adjusts the load peak by discharging during these periods. In contrast, during the period of low electricity prices, DERs operate in the opposite state. It can be seen that the peak-shaving behavior of DERs is within the upper and lower limits, which shows that the customized output allocation method can fully consider the operating characteristics of DERs.

[0241] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market, characterized in that the steps include: Establish a multi-market model based on the power supply network with participation of virtual power plants; The multi-market model includes at least an energy arbitrage market model and a distributed network operator market model, and calculates the expected return and risk prediction value of a single participant in the multi-market model; According to the portfolio method, the combined expected return of the multi-market investment portfolio is calculated by the expected return, and the combined risk value of the multi-market investment portfolio is calculated by the risk prediction value; Based on the risk and return optimization model, the expected return of the portfolio and the risk value of the portfolio are brought in to obtain the optimal investment portfolio; the optimal bidding strategy is obtained through the optimal investment portfolio; Obtaining a clearing result of the virtual power plant based on the optimal bidding strategy; Establish a multi-level efficiency distribution model, which at least includes an output distribution model and a benefit distribution model; The output allocation model allocates the clearing results of the virtual power plant to the distributed energy cluster, and calculates the optimal output allocation strategy of the virtual power plant to the distributed energy cluster through a dynamic stochastic control model and a mean field control framework; The profit distribution model distributes the profit of the virtual power plant based on the Shapley value method, the risk value method and the optimal output allocation strategy.

2. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1 is characterized in that: The expected income of a single participant in the energy arbitrage market model is the sum of the expected unit income multiplied by the number of arbitrage units in the settlement period, and the formula is: In the formula, R E is the expected return of a single participant, EAM is the number of arbitrage units, r EAM,t is the expected unit revenue; n is the number of settlement cycles, and t is the settlement cycle; The formula for the expected unit revenue is: In the formula, is the price of electricity in settlement period t, is the purchase price of electric power in the settlement period t; The risk prediction value of a single participant in the energy arbitrage market model is the standard deviation of the predicted value of each energy sale price, and the formula is: In the formula, σ EAM,t is the risk prediction value of a single participant in the energy arbitrage market model, is the xth of the n predicted values ​​of the selling price in settlement period t.

3. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 2 is characterized in that: The distributed network operator market model establishes branches and nodes based on congestion management technology; The expected profit of a single participant in the distributed network operator market model is the sum of the expected unit price difference multiplied by the number of transaction units during the settlement period, and the formula is: In the formula, R D is the expected return of a single participant, DNO is the number of trading units, r DNO,t is the expected unit price difference; n is the number of settlement cycles, and t is the settlement cycle; The formula for the expected unit price difference is: In the formula, CC l,t is the congestion cost of branch l in settlement period t, the cost of purchasing energy for the distributed network operator market; The risk prediction value of a single participant in the distributed network operator market model is the standard deviation of the congestion volume in each time period, and the formula is: In the formula, σ DNO,t is the risk prediction value of a single participant in the distributed network operator market model, CC i,t,z is the zth of the h predicted values ​​of the congestion cost due to load changes in settlement period t.

4. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1, characterized in that: The multi-market model also includes a frequency modulation market model; The expected income of a single participant in the frequency modulation market model is the sum of the expected income of the frequency modulation service unit in the settlement period multiplied by the number of frequency modulation service units, and the formula is: In the formula, R F is the expected revenue of a single participant in the frequency modulation market model, FRM is the number of frequency modulation service units, r FRM,t is the expected revenue of the frequency modulation service unit; n is the number of settlement cycles, and t is the settlement cycle; The expected revenue per frequency regulation service unit is the unit price of the frequency regulation service and the unit energy cost of the frequency regulation service, and the formula is: Where, REF t is the unit price of frequency modulation service, Unit energy cost for frequency regulation services.

5. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 4 is characterized in that: The process of calculating the unit price of frequency modulation service of a single participant in the frequency modulation market model includes: Calculate capacity benefits and capacity response benefits; The capacity benefit is calculated as: AF=LF+HF=LFR×LA+HFR×HA; Wherein, AF is the capacity benefit of a single participant, LF is the adjustable capacity fee, HF is the adjustable capacity fee, LFR is the adjustable capacity unit fee, HFR is the adjustable capacity unit fee, LA is the adjustable capacity of the virtual power plant within the time period, and HA is the adjustable capacity of the virtual power plant within the time period; The formula for the capacity response benefit is: RE t =CAP t ×RP t ; In the formula, RE t is the capacity response revenue in settlement period t, CAP t is the response capacity of settlement period t, RP t Pay a price for the unit response; The formula for the unit response payment price is: Where, E is the number of capacity frequency modulation, MIP e,t MIV is the market price index for the e-th capacity modulation in settlement period t. e,t is the market volume index for the e-th capacity adjustment in settlement period t; FX t To pay the price floor; The formula for the unit price of the frequency modulation service is: REF t =AF+RE t =AF+CAP t ×RP t ; Where, REF t The unit price for frequency modulation service.

6. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 4 or 5, characterized in that: The risk prediction value of a single participant in the frequency modulation market model is the standard deviation of the energy price when the virtual power plant purchases energy from the system, and the formula is: In the formula, σ FRM,t is the risk prediction value of a single participant in the FM market model, is the yth of m predicted values ​​of the energy price for frequency modulation in settlement period t.

7. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1, characterized in that: The expected return of the portfolio is the sum of the weighted profitability of each market share, calculated as: In the formula, R mp,t is the expected return of the portfolio, ω i is the weight of the virtual power plant in the portfolio in each market, r i,t is the unit revenue of the virtual power plant in the settlement period t of the i-th market, P VPP,t is the total output of the virtual power plant in settlement period t, R i,t is the profit of the virtual power plant when the output is 100% in the settlement period t of the i-th market; M is the market set; The portfolio risk value is calculated by the sum of the risk prediction values ​​of the single participants of each market share in the portfolio and the correlation between any two markets, and the formula is: In the formula, σ mp,t is the portfolio risk value in settlement period t, ω i is the weight of the virtual power plant in the portfolio in the i-th market, ω j is the weight of the virtual power plant in the portfolio in the jth market; σ i,t is the risk prediction value of the ith market in settlement cycle t, σ j,t is the risk prediction value of the jth market in settlement cycle t; ρ ij is the correlation coefficient between the ith market and the jth market.

8. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1, characterized in that: The calculation process of the risk and return optimization model includes: Establish risk minimization models and profit maximization models; Establishing a utility function based on the results of the risk minimization model and the benefit maximization model; The utility function is solved by Lagrangian function, and the optimal investment portfolio is obtained.

9. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1, characterized in that: The output distribution model calculation process includes: Establishing a set of participants; Construct decision vector space; Establishing a utility function based on the set of participants and the decision vector space, wherein the utility function includes a preference degree parameter in the interaction of the participants; Establishing the dynamic random control model through the utility function and incentive strategy; The mean field control framework is established based on the dynamic random control model, and the optimal output allocation strategy is solved.

10. The multi-level allocation method for the performance of a virtual power plant participating in a multi-coupling market according to claim 1, characterized in that: The profit distribution model calculation process includes: Establish the profit sharing function of the participants based on the Shapley value method; Based on the risk utility of distributed resources, the output deviation of distributed power sources and the revenue contribution of operators, the revenue sharing function is optimized to obtain an improved revenue sharing function; The profit distribution strategy of the virtual power plant is obtained by solving the improved profit sharing function.

Images